Attention
Bistable Figures Task
Also called: Necker cube task, Necker cube test, reversible figures, ambiguous figures task, multistable perception, perceptual reversals, reversal rate
Measure perceptual switching: how often a Necker cube flips between its two views, how long each view lasts, and how far people can hold one view on purpose.
8-10 minTimed: desktop or laptop with a mouse or trackpad
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 97
- Evidence grade
- C
- Registered clinical studies
- 2
What it measures
A measure of perceptual switching. Participants watch a Necker cube, a drawing that can be seen from above or from below, and press an arrow key at each change of view: first freely, then while trying to hold one view. Scores are the switch rate, how long each view lasts and how much holding shifts the time spent in the held view. A higher switch rate is not a better score.
Constructs
- Perceptual switching
- Voluntary control of perception
- Viewing-from-above bias
Research uses
- Perception and consciousness research
- Autism, schizophrenia and mood disorder research
- Individual differences research
How the paradigm works
The Bistable Figures Task measures perceptual switching: how often a figure that can be seen in two ways changes from one view to the other while the picture stays the same, and how far a person can hold one view on purpose. The figure is a Necker cube (Necker, 1832), a line drawing of a cube in which either square face can be seen as the front. Seen from above, the front face is at the lower left; seen from below, it is at the upper right. Participants press the left arrow key for the view from above and the right arrow key for the view from below, at the start of each viewing and at every change, with their eyes on a small cross in the middle of the cube, as in Toppino (2003) and Kornmeier and colleagues (2017).
More about the paradigm
The standard session has 2 viewings of 60 seconds in which the participant just watches, then one viewing spent trying to hold each view, in a seeded order (Toppino, 2003; Meng and Tong, 2004). Free viewing comes first, so that no instruction to hold carries into it. The length of the viewings (30 to 90 seconds), the free viewings (1 to 4) and the held viewings for each view (0 to 2) are settings. Before them come a 30-second practice with a cube that really changes between the two shaded views, with feedback on each key, and a 30-second practice viewing of the drawing. A closing 30-second check with a cube that really changes, without feedback, shows whether the keys still named the views. Every key press is saved with its time from the figure's first frame. About 8 to 10 minutes; desktop and laptop only.
The main score is the switch rate: changes of view per minute of free viewing. The mean and median time each view lasts, the share of time the cube is seen from above (most people see it from above more often), the switch rate while holding and how much holding raises the time spent in the held view are also reported. Trying to hold a view mainly lengthens the time spent in it and changes the number of reversals less (Liebert and Burk, 1985; Toppino, 2003). Switch rates differ widely between people (Cao and colleagues, 2018). Scores support research and do not diagnose.
Scores
- Switch Rateswitches/min
- Changes of reported view per minute of free viewing.
- Neither higher nor lower is better; people differ widely
- Mean View Durationseconds
- Mean time a view lasted in free viewing, over the views that began and ended with a change; the first and last view of each viewing are left out. Null with fewer than 3.
- Longer means slower switching
- Median View Durationseconds
- Median time a view lasted in free viewing, over the same views as the mean. Null with fewer than 3.
- Longer means slower switching
- Complete Viewscount
- Views in free viewing that began and ended with a change, which the durations use.
- Informational
- Share Seen From Above%
- Share of the reported time of free viewing in which the cube was seen from above.
- Above 0.5 is the common bias to see the cube from above
- Gain From Holding%
- How much trying to hold a view raised the share of time spent in it, in percentage points over its share in free viewing, averaged over the two views.
- Positive values mean holding worked; larger values mean more voluntary control
- Share Held From Above%
- Share of the reported time in the view from above while trying to hold it.
- Compare with the share in free viewing
- Share Held From Below%
- Share of the reported time in the view from below while trying to hold it.
- Compare with the share in free viewing
- Switch Rate While Holdingswitches/min
- Changes of reported view per minute while trying to hold a view.
- Informational
- Change in Switch Rate When Holdingswitches/min
- Switch rate while holding minus the switch rate in free viewing.
- Negative values mean fewer changes while holding
- Mean Duration of the Held Viewseconds
- Mean time the held view lasted while holding, over views that began and ended with a change. Null with fewer than 3.
- Compare with the other view
- Mean Duration of the Other Viewseconds
- Mean time the view not being held lasted while holding, over views that began and ended with a change. Null with fewer than 3.
- Compare with the held view
- Closing Check Accuracy%
- Share of the views in the closing check, with a cube that really changes, named with the right key within 2 seconds. Null with fewer than 4 views.
- Under 50% is invalid and under 75% is flagged for review
- Closing Check Report Timems
- Median time from a change of the cube in the closing check to the right key.
- Informational; the delay before each reported change
- Free Viewings Without a Changecount
- Free viewings in which no change of view was reported.
- No change in any is flagged for review
- Time to First Reportms
- Median time from the cube's appearance to the first key of a viewing.
- Informational
- Repeated Keyscount
- Presses of the key for the view already reported, which change nothing.
- Informational
- Slipscount
- Reports reversed within 300 ms, which are dropped.
- Informational
- Anticipationscount
- Keys under 100 ms after the cube appeared or after a change in the closing check. They are not taken as reports.
- Informational
- Free Viewing Timeseconds
- Measured time of the free viewings.
- Informational
- Completed Viewingscount
- Viewing periods in the scored run.
- Fewer than planned is flagged for review
- Planned Viewingscount
- Viewing periods the run set out to show.
- Informational
Use and evidence
97 papers from 2021 to 2025 name this task in their title or abstract, down 26% compared with 2016 to 2020. That ranks it 189 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 16 |
| 2024 | 11 |
| 2023 | 23 |
| 2022 | 25 |
| 2021 | 22 |
| 2020 | 33 |
| 2019 | 25 |
| 2018 | 23 |
| 2017 | 25 |
| 2016 | 25 |
| 2015 | 24 |
| 2014 | 23 |
| 2013 | 22 |
| 2012 | 29 |
| 2011 | 31 |
| 2010 | 54 |
| 2009 | 16 |
| 2008 | 22 |
| 2007 | 22 |
| 2006 | 23 |
| 2005 | 20 |
| 2004 | 21 |
| 2003 | 13 |
| 2002 | 3 |
| 2001 | 8 |
| 2000 | 11 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 1
- Papers on reliability or validity
- 32
- Registered clinical studies using it as an outcome
- 2
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.14.0, counted 2 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Figure
- A Necker cube on a dark panel: two squares 160 px wide, the back one 64 px up and to the right, joined at the corners, with a small cross in the middle. Seen from above, the front face is at the lower left; seen from below, it is at the upper right.
- Answer
- Press the left arrow for the view from above and the right arrow for the view from below, at the start of each viewing and at every change. A line under the cube names the two keys.
- Viewings
- 2 free viewings of 60 seconds (standard; 1 to 4 free viewings of 30 to 90 seconds), then 1 viewing spent trying to hold each view (standard; 0 to 2 for each), the view held first chosen by the seed. Free viewing comes first. Each viewing starts with the cross alone for 1,000 ms.
- Practice
- A 30-second practice with a cube that really changes between the two shaded views every 2.5 to 4.5 seconds, with feedback on each key, then a 30-second practice viewing of the drawing.
- Check
- A closing 30-second check with a cube that really changes, without feedback. A change counts as reported when the first arrow key within 2 seconds names the new view.
- Reports
- A key under 100 ms after the cube appears is an anticipation, a key for the view already reported changes nothing, and a report reversed within 300 ms is a slip and is dropped.
- Score
- Changes of view per minute of free viewing. View durations use only the views that began and ended with a change.
Toppino had observers hold a button down while they saw one view; this version, like Kornmeier and colleagues, gives each view its own key. The cube and its shaded views are drawn by our code. Gaze is not tracked: participants are asked to keep their eyes on the cross.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: bistable_figures_test from the scored run, one per viewing period, and the closing check with a cube that really changes. A viewing has no correct answer, so accuracy is not judged against chance. A viewing with no arrow key at all is a missed response: more than max(3, 25%) such viewings invalidate and any is flagged. Naming under 50% of the closing check's views right, with at least 4 views, invalidates and under 75% is flagged. No change of view in free viewing, a run short of its planned viewings or a closing check with no score is flagged for review.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Each viewing asks for the view seen at the start, so a viewing with no key at all means the participant was not reporting; the timed-test rule for missed responses applies. In the closing check the cube itself changes, so the right key is known: wrong keys there mean the keys were swapped or the task was not followed. Some people rarely see a reversible figure flip (Kornmeier and colleagues, 2017), so no change in free viewing is flagged, not invalid. A hidden tab stops the display and the timing of the keys, so a percentile is withheld after one.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Viewing length (seconds) | 60 | 30 to 90 |
| Free viewings | 2 | 1 to 4 |
| Held viewings for each view | 1 | 0 to 2 |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Timed test: participants need a desktop or laptop with a mouse or trackpad. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Bistable Figures Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Necker, 1832; Toppino, 2003), with the standard settings (viewing length (seconds): 60; free viewings: 2; held viewings for each view: 1). The main score was switch rate (changes of reported view per minute of free viewing).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Necker, L. A. (1832). Observations on some remarkable optical phaenomena seen in Switzerland; and on an optical phaenomenon which occurs on viewing a figure of a crystal or geometrical solid. The London and Edinburgh Philosophical Magazine and Journal of Science, 1(5), 329-337. https://doi.org/10.1080/14786443208647909The figure.
- Toppino, T. C. (2003). Reversible-figure perception: Mechanisms of intentional control. Perception & Psychophysics, 65(8), 1285-1295. https://doi.org/10.3758/BF03194852One-minute viewings of the cube reported throughout, with and without the intention to hold one view.
- Meng, M., & Tong, F. (2004). Can attention selectively bias bistable perception? Differences between binocular rivalry and ambiguous figures. Journal of Vision, 4(7), 2. https://doi.org/10.1167/4.7.2Strong selective control of the Necker cube's views.
- Liebert, R. M., & Burk, B. (1985). Voluntary control of reversible figures. Perceptual and Motor Skills, 61(3), 1307-1310. https://doi.org/10.2466/pms.1985.61.3f.1307Holding a view changes the time spent in it more than the number of reversals.
- Kornmeier, J., Wörner, R., Riedel, A., & Tebartz van Elst, L. (2017). A different view on the Necker cube: Differences in multistable perception dynamics between Asperger and non-Asperger observers. PLOS ONE, 12(12), e0189197. https://doi.org/10.1371/journal.pone.0189197The views from above and from below reported with two keys, and the 300 ms floor for a view.
- Cao, T., Wang, L., Sun, Z., Engel, S. A., & He, S. (2018). The independent and shared mechanisms of intrinsic brain dynamics: Insights from bistable perception. Frontiers in Psychology, 9, 589. https://doi.org/10.3389/fpsyg.2018.00589One-minute viewings with two keys; switch rates differ widely between people.
- Troje, N. F., & McAdam, M. (2010). The viewing-from-above bias and the silhouette illusion. i-Perception, 1(3), 143-148. https://doi.org/10.1068/i0408The bias to see a figure from above.
Source papers
- Toppino TC (2003). Reversible-figure perception: mechanisms of intentional control. Perception & Psychophysics, 65(8):1285-1295.Paradigm source, Adults
Questions
How long does the Bistable Figures Task take?
About 8 to 10 minutes, as listed in the test library.
What device does a participant need?
Timed test: participants need a desktop or laptop with a mouse or trackpad. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Switch Rate, Mean View Duration, Median View Duration, Complete Views, Share Seen From Above, Gain From Holding, Share Held From Above, Share Held From Below, Switch Rate While Holding, Change in Switch Rate When Holding, Mean Duration of the Held View, Mean Duration of the Other View, Closing Check Accuracy, Closing Check Report Time, Free Viewings Without a Change, Time to First Report, Repeated Keys, Slips, Anticipations, Free Viewing Time, Completed Viewings and Planned Viewings. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Bistable Figures Task used in research?
97 papers from 2021 to 2025 name it in their title or abstract, down 26% compared with 2016 to 2020. That ranks it 189 of 395 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.14.0, counted 2 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Bistable Figures Task have?
In the published research, 1 paper reports its test-retest reliability and 2 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Viewing length (seconds) (30 to 90), free viewings (1 to 4) and held viewings for each view (0 to 2). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.